Background: Blood transfusions save lives and improve health; however, unnecessary transfusion practice exposes patients to immediate and long-term negative consequences. One of the 2013 Choosing Wisely®recommendations focused on avoiding liberal red blood cell (RBC) transfusion. The specific recommendation was that in situations where transfusion of RBCs is necessary, transfuse the minimum number of units required to relieve symptoms of anemia or to return the patient to a safe hemoglobin (Hb) range (7-8 gr/dl in stable, non-cardiac inpatients). As part of an initiative to improve the quality of RBC transfusion practice we performed this study to examine RBC transfusion practices by patient- and admission-related characteristics.Methods: A cross-sectional survey of RBC transfusions was conducted at three Jerusalem hospitals in the following departments: intensive care (general, adult), internal medicine, orthopedics, hematology and cardiothoracic surgery. These departments were chosen based on their high volume of RBC use. RBC transfusions were identified from the list of RBC units discharged from the hospital’s blood bank within a specified 30-day period and data were collected daily by trained nurses. Exclusion criteria included patients who underwent liver transplants, patients with solid tumors, patients requiring massive transfusions (≥6) and transfusions given during surgery. Orthopedic and cardiothoracic departments were grouped together as surgical departments and compared to all others (i.e. non-surgical). Off-protocol RBC transfusion was defined in this study as patients receiving>1 RBC unit consecutively or transfusion given to non-bleeding, non-active cardiac patients with Hb levels ≥8 gr/dl. Generalized estimating equations (GEE) method was applied to assess the separate associations of each selected characteristic with off-protocol RBC transfusion, taking into account clustering of observations due to repeated transfusions per patient.Results: During the study period 584 RBC transfusions met inclusion criteria. These transfusions were given to 302 patients, mean age of 67 (±19.5, range 14 to 100) years, of which 162 (53.6%) were female. Mean number of transfusions per patient was 1.9 (±1.3) [range from 1 (in 52.6%) to 8 (in 2) patients]. Nearly all patients (291, 96.4%) had at least one underlying medical condition; hypertension (48.3%), malignancy (33.1%), heart (44.7%), nephrology (32.1%) and pulmonary diseases (32.1%). Antithrombotic therapy was taken by 142 (47%) patients. Of the 584 RBC transfusions, 498 (85.3%) were given in the non-surgical departments and 247 (42.3%) were given to patients who underwent invasive procedures/surgery during the current admission, of which 137 (55.5%) were considered major operations (e.g. open laparotomy, open heart, etc.). Pre-transfusion Hb level was ≥8 gr/dl in 229 (39.2%) transfusions and>1 RBC unit was given consecutively in 96 (16.4%) transfusions. The prevalence of off-protocol RBC transfusion, as defined in this study, was 48.1%. Mean age of patients receiving off-protocol RBC transfusion was higher than those receiving by protocol (67.8±18.2 vs. 60.7±21.8, OR=1.02, 95% CI 1.01-1.03). Off-protocol RBC transfusion was more common in the surgical departments vs. non-surgical (OR=7.4, 95% CI 3.7-14.7). In patients undergoing invasive procedure/surgery, major operations were associated with higher odds of off-protocol RBC transfusion compared to minor procedures (OR=1.7, 95% CI 1.1-2.8). Off-protocol RBC transfusion was not related to presence of underling malignancy, heart, nephrology and pulmonary diseases, but was more common among patients taking antithrombotic therapy (OR=1.7, 95% CI 1.2-2.4). Pre-transfusion recording of patients’ blood pressure, pulse rate and saturation were not associated with off-protocol RBC transfusion.Conclusions: This study demonstrates that almost half of RBC transfusions are not given based on suggested guidelines. Although clinical considerations, such as underlying diseases or patient's pre-transfusion signs, may explain non-adherence to guidelines, no clear pattern was observed in the current study to support this explanation. The study findings highlight the need to further our understanding of clinical decision making leading to RBC transfusion and call for establishing clear guidelines to facilitate wise transfusion-related choices. DisclosuresNo relevant conflicts of interest to declare.